State-of-the-Art and Future Directions in Autonomous Navigation for UAVs in GNSS-Denied Environments: A Comprehensive Review of Techniques, Architectures, and Applications

Al-Qadası, YUNES; Alamerı, Abdullah; Abdallah, Ayman; Ahmed, Gamil

doi:10.1016/j.inffus.2026.104714

State-of-the-Art and Future Directions in Autonomous Navigation for UAVs in GNSS-Denied Environments: A Comprehensive Review of Techniques, Architectures, and Applications

Al-Qadası Y. S. N., Alamerı A., Abdallah A. M., Ahmed G.

INFORMATION FUSION, sa.104714, ss.1-42, 2026 (Scopus)

Yayın Türü: Makale / Tam Makale
Basım Tarihi: 2026
Doi Numarası: 10.1016/j.inffus.2026.104714
Dergi Adı: INFORMATION FUSION
Derginin Tarandığı İndeksler: Applied Science & Technology Source, Academic Search Ultimate (EBSCO), Engineering Source (EBSCO), Scopus, Compendex, INSPEC
Sayfa Sayıları: ss.1-42
Kocaeli Üniversitesi Adresli: Evet

Özet

The widespread deployment of Unmanned Aerial Vehicles (UAVs) in complex operational settings remains constrained by reliance on Global Navigation Satellite Systems (GNSS) for accurate positioning and attitude stabilization. This manuscript provides a comprehensive review of state-of-the-art methods and architectural frameworks that enable robust autonomous navigation in GNSS-denied, degraded, or jammed environments, including indoor spaces, underground tunnels, and dense urban canyons. The review establishes a clear taxonomy that classifies navigation strategies into relative localization methods (e.g., odometry and Simultaneous Localization and Mapping, SLAM) and global localization methods based on external references and prior maps. A comparative synthesis is presented across core sensing modalities, visual, LiDAR, and inertial, highlighting their strengths, limitations, and their integration within high-performance fusion pipelines such as Visual–Inertial Odometry (VIO) and LiDAR–inertial systems. The article further surveys advanced approaches, including Ultra-Wideband (UWB) localization, Terrain-Aided Navigation (TAN), radar-based navigation, and Deep Learning and Reinforcement Learning (DRL), and discusses their roles within integrated and cooperative navigation architectures. Key operational challenges are examined, with emphasis on Size, Weight, and Power (SWaP) constraints, environmental variability, failure detection and recovery, and the long-term drift that can arise in relative methods without absolute correction. Benchmark-oriented comparisons are also synthesized to clarify reported performance trends and practical trade-offs across representative navigation frameworks. The manuscript concludes by outlining research directions focused on efficient learning models, reliable multi-modal fusion, semantic-aware mapping, and certifiable resilience mechanisms to support safety-critical GNSS-denied missions such as infrastructure inspection, search and rescue, and environmental monitoring, with relevance to both research and industrial development.